This is causing a lot of tryjobs to fail recently (29 in the last 24 hours).

I believe https://codereview.chromium.org/2624523003 is the cause, given that it ladned but also had this failure in it and touched this code.

looks like it was reverted though. please unflake the test before relanding it

I wonder if this was caused by https://codereview.chromium.org/2609713002/ instead.  That CL landed around 18:50 UTC, and all but one of the flakes started around 19:35 UTC

The flakes appear to have the following error:

[ERROR:connection.cc(1947)] OfflinePageMetadata sqlite error 1802, errno 0: disk I/O error, sql: CREATE TABLE IF NOT EXISTS offlinepages_v1(offline_id INTEGER PRIMARY KEY NOT NULL, creation_time INTEGER NOT NULL, file_size INTEGER NOT NULL, last_access_time INTEGER NOT NULL, access_count INTEGER NOT NULL, client_namespace VARCHAR NOT NULL, client_id VARCHAR NOT NULL, online_url VARCHAR NOT NULL, file_path VARCHAR NOT NULL, title VARCHAR NOT NULL DEFAULT '', original_url VARCHAR NOT NULL DEFAULT '')
[FATAL:connection.cc(1962)] disk I/O error

which may be something related to no longer delaying the OfflinePageModel construction when creating the test RequestCoordinator...

Prognosis is that the RequestCoordinatorFactoryTest doesn't flush the i/o thread (only flushes browser thread), so causing the flake depending on which order the threads terminate. Will test locally

This test is still flakey, and I seem to be hitting it on every run!

This is blocking the entire Chromium commit queue for changes which affect the build, because these aren't retried without the patch. Examples:

https://build.chromium.org/p/tryserver.chromium.android/builders/android_n5x_swarming_rel/builds/97761
https://build.chromium.org/p/tryserver.chromium.android/builders/android_n5x_swarming_rel/builds/97738

Please disable this test right away while the flakiness is being investigated. Thanks.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/95ae77193e1dbd5a7e6048095d0100597e626e83

commit 95ae77193e1dbd5a7e6048095d0100597e626e83
Author: Philip Jägenstedt <foolip@chromium.org>
Date: Wed Jan 11 15:46:56 2017

Disable RequestCoordinatorFactoryTest.BuildRequestCoordinator (flaky)

BUG= 679844 
TBR=petewil@chromium.org

Review-Url: https://codereview.chromium.org/2622303002 .
Cr-Commit-Position: refs/heads/master@{#442911}

[modify] https://crrev.com/95ae77193e1dbd5a7e6048095d0100597e626e83/chrome/browser/android/offline_pages/request_coordinator_factory_unittest.cc

Detected 180 new flakes for test/step "RequestCoordinatorFactoryTest.BuildRequestCoordinator". To see the actual flakes, please visit https://chromium-try-flakes.appspot.com/all_flake_occurrences?key=ahVzfmNocm9taXVtLXRyeS1mbGFrZXNyQAsSBUZsYWtlIjVSZXF1ZXN0Q29vcmRpbmF0b3JGYWN0b3J5VGVzdC5CdWlsZFJlcXVlc3RDb29yZGluYXRvcgw. This message was posted automatically by the chromium-try-flakes app. Since flakiness is ongoing, the issue was moved back into Sheriff Bug Queue (unless already there).

After a few days of debugging and trying things, we are going to remove the test because...

1. The test as it currently stands isn't adding much value in the first place
2. After many efforts described below, the test is still flaky
3. I have yet to see another keyservice factory with a unittest
4. This investigation is turning to be too long with little return value

Steps taken to try to solve the flake:
1. Since the issue was always a disk I/O error, tried flushing all the threads, including the IO thread. Turned into consistently failing.
2. Added a DependsOn between the RequestCoordinatorFactory and the OfflinePageModelFactory, which was implicit before but now more explicit. Did nothing regarding the flake
3. Rather than extending from a render harness, create our own thread bundle, profile manager and testing profile. This had marginal effect. Rather than having just a disk I/O error, we now also have "unable to open database file"... but we also have a 50% flake rate.

I will send CL to include the DependsOn change and to delete this test shortly.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f109c99563e6efe928faee2ba68402f1a3cc06f3

commit f109c99563e6efe928faee2ba68402f1a3cc06f3
Author: chili <chili@chromium.org>
Date: Wed Jan 18 05:36:13 2017

[Offline Pages] Add explicit DependsOn clause to request coordinator factory
and delete a flaky test

BUG= 679844 

Review-Url: https://codereview.chromium.org/2635633002
Cr-Commit-Position: refs/heads/master@{#444285}

[modify] https://crrev.com/f109c99563e6efe928faee2ba68402f1a3cc06f3/chrome/browser/android/offline_pages/request_coordinator_factory.cc
[delete] https://crrev.com/1f89b6d7e48c06aec8f4e2c0723afa229045a402/chrome/browser/android/offline_pages/request_coordinator_factory_unittest.cc
[modify] https://crrev.com/f109c99563e6efe928faee2ba68402f1a3cc06f3/chrome/test/BUILD.gn

+fgorski, who expressed interest in looking at the broken unittest for science